claim
active
claim:no-suppressive-mechanism-prevents-answering-with-entities-absent-from-contextNo suppressive mechanism prevents answering with entities absent from context
Rules out alternative explanation.
Source paper
extracted_from(2025) · Yoav Gur-Arieh · Mor Geva · Atticus Geiger
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Patching an unresolvable pointer at layer ℓ yields no counterfactual answer, but at ℓ+1 yields itsupportsValidates reflexive pointer.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- When lexical/reflexive targets absent from original, model falls back to noisy positional mechanismfinding0.740Fig. 28–29.
- Generalizes beyond multiple choice.
- Explains why the boundary appears fixed: the prior hides its context-dependent nature.
- About chain-of-thought and process safety.
- Applied security implication derived from the asymmetry finding.
- Architectural explanation.
- noted as a possible confound
- Cited regarding steganographic encoding of reasoning in chain-of-thought